add async copy for openfold swa triton kernel

NVIDIA / apex

A PyTorch Extension: Tools for easy mixed precision and distributed training in Pytorch

BSD 3-Clause "New" or "Revised" License

8.17k stars 1.35k forks source link

Closed azazhu closed 6 months ago

azazhu commented 7 months ago

This MR is to reduce cpu-gpu interaction for openfold